Tuesday, December 28, 2010

The Best Offensive Teams in the Big Leagues: The 2010-11 Season To Date

Here's another "end of the year" post about where we stand halfway through a very exciting season in the Big 4 leagues of international soccer (Bundesliga, EPL, La Liga, and Serie A).* Below is a set of calculations of offensive production (goals per match) at home (the first graph) and away (the second graph). The graphs show how the teams ranked from most to fewest goals scored as of Dec. 21.

Aside from the fact that the distribution of goals scored at home is slightly to the right compared to what it is when teams play away (reflecting the fact that teams, on average, score more at home than away), we see a huge range of goals scored across teams in the four leagues. So take a look to see who's been great and who's been woeful offensively. First, here are the home goal scoring records of all teams.

The best of the best in Europe's top leagues when playing at home is clearly Real Madrid with an impressive 3.25 goals per match; in fact, it's not even close. Overall, the Top 3 teams are Real, Man U, and Barcelona, with the latter two teams scoring slightly fewer than 3 goals when playing at home. And while these three teams are in a league by themselves, a good number of teams score more than 2 goals per match when they play at home. The Top 10 home scoring teams in Europe also include Stuttgart, Juventus, Arsenal, Villareal, Dortmund, Chelsea, and Bayern Munich.

So who's the worst offensive team at home? Genoa "lead" the pack of teams that cannot score at home with .75 goals per home match. And four other teams have scored fewer than one goal per home match: Bari, Parma, Cesena, and Wigan - catenaccio anyone? - revealing that bad Serie A teams just don't score. The biggest surprise among the low-scoring teams: Man City with 1 goal per home match (and this is not counting its most recent home loss).

So what happens to offensive production when teams travel? Take a look at s goals scored on the road.

Here we see that Barca is playing by far the best offense on the road with an astonishing 3.625 goals per match when playing away from Camp Nou. But the interesting thing about the numbers is not simply that no other team comes close, but that the next four teams are all from the Bundesliga, with Leverkusen leading the pack at 2.4 goals, followed by Dortmund (2.3), Mainz (2.1), and lowly Moenchengladbach (1.9). Interestingly, Man City has been scoring much more on the road than at home.

On the flip side, lots of teams score fewer than 1 goals when playing away, some even fewer than half a goal. The bottom 5 teams all score fewer than .5 goals when playing away from home, and again, none of them are Bundesliga teams. Instead, they include Lecce, Brescia, and Cesena (again) from Serie A, and La Coruna and Hercules from La Liga. Hercules' average of .375 road goals their win at Barca's Camp Noú on Sep.11 even more remarkable than it otherwise already was.

Comparing the leagues, these numbers suggest that offensive production is very lopsided in La Liga, with the best teams shooting their opponents' lights out, while the worst teams can't buy a goal (to mix metaphors just a bit). And while things aren't quite as uneven in Serie A, they're close. So, on offensive production alone, Bundesliga and Premier League teams appear more evenly matched so far this season.

In the next few days, I'll take a look at defensive production, too, to see where we stand halfway through the season.

* With apologies to all the hardworking soccer professionals around the globe. In the new year, I hope to have more posts comparing more leagues than "just" the Big 4. Stay tuned.

Friday, December 24, 2010

SQ's Harebrained Soccer Theory of the Year: The Nutella Curse and the Curse of the Nike Ad

World Cup years bring all kinds of interesting theories about what explains performance in soccer, and this year was no exception. This year I found two of them particularly entertaining - in a ludicrous kind of way. Because they are essentially the same theory, they share SoccerQuantified's harebrained theory of the year award: they're the Nutella Curse and the Curse of the Nike Ad.

The story goes something like this: It all starts very innocently. Young, handsome soccer players are signed up (probably by the dark forces of marketing brand management) to endorse a product. Unbeknownst to them, signing up with this particular product unleashes the dark curse associated with the product and their soccer career subsequently suffers immensely. Some of them are never heard from again.

Don't believe me? Well, there's lots of "evidence" to back up the theory. Since roughly 2004, Ferrero-Rocher has been using young, up and coming German soccer players to advertise Nutella (a chocolate-hazelnut spread). Hardly any of them have made it big and their once-promising careers have taken turns for the worse. Nutella-cursed hopefuls include Kevin Kuranyi (who went from being the Bundesliga's top scorer to being dumped by the German national team and then sold by Schalke to Dynamo Moscow), Benny Lauth (who? allegedly, Germany's best striker since Gerd Müller), Andreas Hinkel, Tim Borowski, Jermaine Jones, Tobias Weis, Simon Rolfes, Marcell Jansen, or Arne Friedrich, none of whom have had easy times ever since endorsing Nutella and several of whom have had long-lasting injuries.

The Curse of the Nike Ad is similar in its set-up. After all, many if not most of the amazing players (Ronaldo, Ribéry, Cannavaro, Rooney, Drogba, etc.) featured in last summer's inspiring and entertaining "Write the Future" ad exited the World Cup early (or didn't even make their national teams - Ronaldinho anyone?). Clearly, endorsing Nike spelled doom for the players' respective national teams (the likes of Italy, France, Portugal, Brazil, or Ivory Coast).

It's pretty simple: endorse the product and get ready for failure on the soccer field. QED.

Of course, nothing could be further from the truth - in case you were actually wondering. These "curses" are classic examples of the old scientific adage that correlation does not imply causation.  It's a simple logical and statistical fallacy.

But back to Nutella. Even if there is a curse, maybe the tide has turned. Take a look at one of the more recent Nutella ads:

It features goalkeeper Manuel Neuer and defender Benedikt Höwedes of Schalke, defender Mats Hummels of Dortmund and Bremen's Mesut Özil. If you ask me, these four have clearly been able to defeat the forces of darkness. And if you pay close attention to the Nike ad, you'll see that Iniesta, Fabregas, and Pique are in it, too - they don't seem too cursed to me.

It's all about selling product and more vodoo than science, but it's still lots of fun.

Thursday, December 23, 2010

Political Violence and Bad Behavior in Soccer: Is There A Connection?

As a great soccer year is coming to a close, I thought I'd start sharing some of the more interesting soccer-related scientific research I've come across. Among the most fascinating papers I've read in the past 12 months is a really creative study by three political economists (Miguel, Saiegh, and Satyanath) about the connection between civil conflict (political violence) in a player's home country "and his propensity to behave violently on the pitch, as measured by yellow and red cards the player received."

The paper's story is straightforward: Since professional footballers now hail from all over the world, many of them come from poorer countries with significant levels of civil strife and political instability, while others were raised in the rich, stable, democratic countries of the West. The authors want to know: Does this matter for how they behave when they play the game? The answer seems to be yes.

Based on data from the 2004–2005 and 2005–2006 seasons in five national leagues (England, France, Germany, Italy, and Spain) as well as the Champions League, Miguel et al. find that civil conflict in a player's home country and affects a player's propensity to behave violently on the soccer field, as measured by yellow and red cards. They arrive at this finding by assigning each player a value for the number of years a country suffered from civil war during the period of time players were presumably growing up in the country and calculating the number of yellow and red cards the player received during the season.

The statistics behind the results will surely bore you to tears if you're not a social scientist, but the effect is nicely captured by a couple of graphs in the paper. Below you can see that, as the number of years a country has experienced civil war goes up, so does the average number of yellow cards per player and season for someone from that country (the bubbles show you how many players are from each of the countries in the study). Take a look.

MIGUEL, E., SAIEGH, S. M. and SATYANATH, S. 2010. "Civil War Exposure and Violence." Economics & Politics (in press).
To quote from the paper:

Worst 11 of the Year So Far: The Premier League Edition

Photograph: Paul Harding/Action Images

Can be found here, courtesy of Paul Doyle at the Guardian. And don't forget to read the comments, too.

Tuesday, December 21, 2010

Guardian Gallery Pic of the Year

Ok, so this belongs in the category of frivolous, waste a few minutes at work, end of the year fun. Since there were so many good Guardian Gallery pictures this year, courtesy of The Guardian newspaper, I couldn't make up my mind which should win the SoccerQuantified pic of the year. So I let my kids be the jury. They loved this one in particular - can you tell who's in the picture? - and it's completely apropos. After all, money is something you actually can put a number on, in soccer and in life.

(c) The Guardian Newspaper 2010

The Worst 11 of the Year: The Bundesliga Edition

When the going got tough, Cologne's Mondragon just left
Ok, so it's that time of year when we look back on things past ...

This Bundesliga Worst 11 list of the year was just too good (or bad) to pass up. Enjoy (and condolences to the Philadelphia Union)!

Sunday, December 19, 2010

And The Winner Is ....: World Cup Bids As Elections

So far as I know there's nothing that requires FIFA's decisions to be democratic or transparent. But in fact, the most recent FIFA Executive Committee decisions to award the 2018 and 2022 World Cups to Russia and Qatar were more transparent and democratic than most people might suspect.

Don't believe me? Check for yourself. To quote FIFA directly, the 2018 and 2022 World Cup bids were decided based on these voting rules:
  • The 2018 vote will take place first, then the 2022 one. The vote will be by secret ballot and all eligible members of the FIFA Executive Committee can vote in both ballot.
  • To win the right to host the competition, a bidder must obtain an absolute majority (50% + 1) of the votes of the FIFA Executive Committee members present.
  • In the event of a tie when only two bidders remain, the FIFA President will have the casting vote
  • For any voting round in which an absolute majority is not achieved, the bidder with the lowest number of votes will not progress to the next voting round
  • If there is a tie for the lowest number of votes in any round, an intermediate voting round will be conducted to determine which of the tied bidders does not progress
  • When the final decision on the host has been taken, the result will be put in two envelopes and taken by the notary to the “Messe Zurich”, where they will be handed over to the FIFA President for the announcements
Here is what we have learned about how the voting actually transpired. In the voting on the 2018 World Cup, the Executive Committee went through two rounds of balloting. England was eliminated in the first round of voting. In Round 2, the Netherlands/Belgium bid lost two votes (going from 4 to 2), and these 2 votes and the 2 England votes went to Russia. As a result, Russia achieved an absolute majority of 13 votes in the second round, beating out the next closest contender Spain/Portugal's.

Below is a graphical representation of the votes.

The committee's voting on the 2022 World Cup took four rounds to decide on the winner. Australia was eliminated in the first round, Japan in the second, and South Korea in the third round of balloting. Qatar won an absolute majority in the 4th round of balloting in a head-to-head matchup against the U.S.

[Here are the really intricate details: So, after Australia was eliminated in the first round, the 1 Australia vote was up for grabs; moreover, in Round 2, Japan lost one vote, Qatar lost one vote, South Korea gained one, and the U.S. gained two votes. In Round 3, South Korea stayed level on votes, and the 2 Japan votes went to the U.S. and Qatar bids, each gaining 1 vote. Finally, in Round 4, with South Korea eliminated and its 5 votes in play, the U.S. gained 2 votes, but Qatar gained 3 votes to achieve an absolute majority of 14 votes.]

So what are we to make of this? Thinking about it from a political science perspective, I think it's worth noting that

Thursday, December 16, 2010

Conceding Halftime Leads: Why Arsenal’s Collapse Against Tottenham Was A One-In-Forty Occurrence

Take a look at this great post from Zach Slaton at A Beautiful Numbers Game, one of my favorite soccermetrics blogs. 

Just how rare was Arsenal’s second half collapse against Tottenham Hotspur on November 20th? Sure it was Tottenham’s first victory at the Emirates, and its first win as the away team in this fixture in seventeen years. Teams come and go in the Premier League as do players, so I don’t put much stock in such streaks. The better question to ask is how rare was this collapse when looking at historical averages. Chris Anderson from Soccer By the Numbers has been kind enough to supply the data behind his recent post on halftime leads and fulltime results (EPL seasons 05/06 through 09/10), and the results do indeed make Arsenal’s collapse look like a very rare event.

First, let’s look at all results without declaring if the team that is ahead at halftime is home or away and whether or not they went into halftime with a clean sheet.  This produces the table below.

The rows on the left indicate the halftime lead, and the columns across the top are the full time result.  The results for a halftime lead of 2 goals indicate that regardless of location or a halftime cleansheet, historical data indicates that Arsenal had a 95% chance of winning, a 3% chance of drawing, and 2% chance of losing.

Taking the analysis one step further, the data set is isolated for leads at home to produce the table below.
In examining this result, Arsenal as the home team still had a 95% chance of winning, a 2.5% chance of drawing, and a 2.5% chance of losing. It turns out that further reducing the data for a cleansheet while at home does not change the numbers for a two goal lead.  Nonetheless, the table below is provided for those interested in other clean sheet halftime leads and their results.
Ultimately, this means Arsenal’s loss was a one-in-forty occurrence given their two goal lead at home with a clean sheet at the half.  Even without Kabul’s winning goal in the 86th minute Arsenal would have still experienced a one-in-twenty collapse in conceding a draw with such a halftime advantage.

Of the other four such collapses the last five seasons, one team features twice and each of the two teams in this latest occurrence play a role in the other two matches.

First, Manchester City experienced such a fate twice in one calendar year.  The first time, April 26th 2008, saw City go ahead 2-0 in the first twenty one minutes, only to concede three goals in the final twenty one minutes of play against Fulham.  Liverpool then repeated the feat of overcoming a 0-2 Man City lead on October 5th, 2008 with Torres and Kuyt keying the Reds’ comeback.

West Ham United experienced a similar collapse against none other than Tottenham on March 4th, 2007.  Carlos Tevez contributed to the 2-0 halftime lead, while Tottenham battled back even by the 63rd minute.  This match was the only one of the four where the team leading at halftime subsequently took the lead again after relinquishing it in the second half, all off a Bobby Zamora goal in the 85th minute as a substitute.  Tottenham then battled back with two goals in the final minutes, including the winner in the 6th minute of stoppage time.

Ironically, Arsenal played the roll of spoiler in the fourth-and-final match as they came back on Bolton on March 29th, 2008.  Not only did the Gunners embarrass Bolton on their home pitch, but they did so down a man the entire second half.

It seems as if Arsenal is in rare company, having been on both ends of such an event.  Tottenahm is also in a rare position, having won two of the five occurrences.  I doubt any of the other collapses were as painful for those teams as Arsenal’s collapse in the latest iteration of the North London Derby.

Monday, December 13, 2010

The Point Value of Yellow Cards in the EPL To Date

A quick follow-up to the earlier post about yellow cards and points earned per match. I thought it might be interesting to see (a) how things are shaping up in the EPL so far this season on this score, and (b) whether there are any home-away team differences worth talking about (given the tendency of home teams to take more points).

So here's how it looks (till the end of November). First, the pattern of yellow cards given so far this year is similar to previous seasons: teams receive no yellow cards in about 20% of matches, and they receive only one in another 26%. The most common occurrence is a team receiving two yellows (35% of matches). In stark contrast, matches with teams receiving 3 or more yellow cards are much less frequent (about 20% of the time).

This relative frequency of yellow cards matters insofar as the vast majority (95%) of matches teams play in involve between 0 and 3 cards. So is there a pattern in how many points teams earn when they "earn" different numbers of yellow cards? Below is the "point value" of a match associated with different numbers of yellow cards, ranging from 0-7 so far this year. Take a look.

Clearly, the graph shows that more yellows are associated with a steadily decreasing chance of earning points. While the point value of zero yellows is 1.5, it is only 1 for a team that received 3 yellow cards in a match. Now the graph also sows that the point value of 4 yellow cards is very high (2), while that of 5 and 6 yellows is zero. It is difficult to get too worked up about these latter patterns, in large part because the number of matches  included in these categories is very small.

Finally, are there home-away team differences? To answer this question, I split up the teams by whether they played at home or away and re-calculated the point values associated with yellow cards. These are shown in the next graph.

Interestingly, the point values of yellow cards are systematically different for home and away teams. For one, home teams that receive no yellow cards on average gain 2 points, while those receiving 1-3 cards on average still take around 1.5 points. In contrast, there is no real pattern in points associated with yellow cards for away teams: they gain right around 1 point on average, almost regardless of how many yellow cards they receive (with the exception of cards in the 4-7 range, a rare occurrence).

Taken together, this curious pattern suggests that the diminishing point value of yellow cards only exists for home teams: the formula fair play = winning play appears to be true only for home teams. Given that referees have not systematically advantaged home teams in their decision so far this season, it seems that away teams don't need to worry all that much about yellow cards when trying to win a match away from home. Whether this is a pattern that will be sustained over the long of a whole season is a question for another post in the new year.

Saturday, December 11, 2010

How Much Do Yellow Cards Hurt A Team? Comparisons Across Leagues

Do yellow cards help or hurt a team's chances to win a match, or do they perhaps not matter at all? The intuitive answer seems to be that they should hurt - that is, if yellow cards are an indicator of a team's inability to cope with the other side's offensive pressures. This would be consistent with the common complaints you hear from managers that teams foul more when they are on the losing end of a match (remember Arsene Wenger's complaints about Stoke and Blackburn?). Of course, it's also possible that yellow cards reflect a team's willingness to play tough defense, and thus might in the end help a team keep the other side at bay.

To get a handle on whether yellow cards help, hurt, or are simply irrelevant, I first took a look at how common yellow cards are across the leagues. The graphs below show the percentages of times a team received a certain number of yellow cards per match for the five seasons stretching from 2005/06 to 2009/10. So, looking at the upper left graph, you can see that Bundesliga teams received 0 yellow cards in about 14% of all matches played during this time.

A couple of other things stand out. The distributions of yellow cards are not wildly different, but they are distinct in a couple of ways. While the distribution in the Bundesliga and the EPL is left-skewed - meaning there are more matches with no or just one yellow card than there are matches with, say, 5 or 6 yellows, the distribution of yellows per team and match in La Liga and Serie A look much more like a normal (bell-shaped) distribution. So, while Bundesliga teams received no yellows in 14% of matches they played and EPL teams received no yellow cards in over 20% of matches they played, a match with no yellow cards is a much less common occurrence in the Spanish and Italian leagues, with a frequency of 6.3% and 8.3% of matches, respectively. In contrast, there were many more matches in which teams saw three cards or more in La Liga and Serie A than in the Bundesliga or the EPL.

So, does it matter for match outcomes - measured in points - if a team receives yellow cards during a match? To answer this question, I calculated the average number of points associated with the number of yellow cards a team received in a match (the "point value", so to speak). Take a look. 

In three of the four leagues, yellow cards are clearly costly for a team's chances to win points. In the Bundesliga, the Premiership, and La Liga, a higher number of yellow cards goes hand in hand with a lower point total. While teams with no yellow cards win slightly more than 1.5 points (1.63 across the leagues to be precise), a team's point value for the match steadily decreases with each additional yellow card.*

But this is clearly not the case in Serie A where there is absolutely no correlation between the number of yellow cards and the points a team wins in that match. Regardless of how many yellow cards a team's players see during a match, the number of points ranges roughly between 1.2 and 1.5. I can think of at least two possibilities why this might be: one, while yellow cards suggest an overwhelmed defense in the the Bundesliga, EPL, and La Liga, they reflect an effective defense in Serie A; or two (and perhaps more plausibly), Serie A referees handle yellow cards differently than referees in the other leagues, perhaps for reasons unrelated to fouls and defensive play. One indicator of this is the Pearson correlation between the number of fouls a team commits in a match and yellow cards given by the referee. It is .37 in the Bundesliga and the Premier League and .32 in La Liga, while it is lowest in Serie A at .28.

So, as you watch your team play this weekend, keep in mind that fair play also means play that brings points. But not in Serie A.

* There is one small outlier to this general pattern, namely the amount of points associated with 6 yellow cards in Bundesliga matches, which is higher than the number of points in matches with 5 cards. The reason has to do with sample size, or the very small number of matches - 11 to be exact - where teams received 6 yellow cards

Wednesday, December 8, 2010

EPL Defensive Production To Date

After taking a look at offensive production, I thought it was time to see where things stand in the Premiership with regard to teams' defensive capabilities. Here are three pieces of simple, but I think informative information (as of Nov.28).

First, very basically, there's the average number of goals teams have allowed per match this season. The numbers show clearly that the best teams in the league are also the best defensive teams. But this is true only for the four best defensive teams (Chelsea, Man City, Man U, and Arsenal). Of these, Chelsea and Man City clearly stand out in allowing fewer than 1 goal(s) per match, followed by Man U and Arsenal who have allowed about 1 goal per match. But perhaps surprisingly, Birmingham, Fulham, and Sunderland, on average, have allowed as few goals as Arsenal and Man U. At the other end of the table, West Brom, West Ham, Wigan, and Wolves have allowed about 1.75 goals per match, and Blackpool takes the cake by allowing almost 2 goals per match.

But these are, of course, just average numbers across a good number of matches, so they hide significant variation over the course of a team's season. To see how common specific defensive scores are for various teams, the second graph below shows the frequency of allowing different numbers of goals. Here, we can see that Chelsea and Man City are the kinds of defenses that succeed at keeping clean sheets, while Bolton and Tottenham have a hard time keeping the ball out of the net entirely, but rarely allow more than one goal.

Finally, a good metric of how important clean sheets and therefore peak defensive performance are to teams' overall fortunes, I calculated the number of points each team has gained so far this year when keeping a clean sheet. Take a look. You can see that clean sheets have had enormous value (3 points, on average!) for West Brom, Stoke, and Arsenal - that is, as of Nov.28, these teams always won their matches when they allowed no goals. In contrast, Tottenham and Bolton have been able to profit, on average, only one point per clean sheet kept. Another interesting twist in the data: when Wigan are able to keep a clean sheet - something that has happened only a few times this year - the club gains 2.5 points (the same is true for Newcastle and Everton, too).

PS: Please remind me to ask someone in charge why goalkeepers are paid less than other positions.

Monday, December 6, 2010

Exit, Voice, and Loyalty: Why FIFA's Monopoly Works As It Does

Last Friday, the day after FIFA announced its decision on the 2018 and 2022 World Cup bids, I had lunch with Andy Markovits, one of the authors of Gaming the World: How Sports Are Reshaping Global Politics and Culture (just named one of the Financial Times' top 3 sports books of the year). Neither of us was particularly happy with the decisions, to be honest, and we talked about how fans might react.

FIFA's global expansion tour, which has taken the tournament to Africa, and now will be taking it to Russia and the Middle East, clearly reflects the organization's global ambition to expand, rather than deepen markets. And China's rumored interest in hosting the 2026 cup would be the next logical step in that expansion.

So what's a fan (or football association, like the English FA) unhappy with FIFA's direction to do? Andy mentioned that you could think of fans' relationships with FIFA in terms of Albert Hirschman’s classic (1970) book Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States. I've been thinking about this over the weekend, so let's spin this out for a second. As Wikipedia explains:
"The basic concept is as follows: members of an organization, whether a business, a nation or any other form of human grouping, have essentially two possible responses when they perceive that the organization is demonstrating a decrease in quality or benefit to the member: they can exit (withdraw from the relationship); or, they can voice (attempt to repair or improve the relationship through communication of the complaint, grievance or proposal for change). For example, the citizens of a country may respond to increasing political repression in two ways: emigrate or protest. Similarly, employees can choose to quit their unpleasant job, or express their concerns in an effort to improve the situation. Disgruntled customers ask for the manager, or they choose to shop elsewhere."
So think of national FA's or fans as the equivalent of FIFA's customers or citizens. If they are unhappy with the decision to award the World Cup to, say, Qatar or the direction of world football generally, they basically have three possible responses. These include:

Saturday, December 4, 2010

Shots and Goals: An EPL Update of Offensive Production Through November

Here's an update on offensive production in the EPL so far this year (using data as of Nov.28). First, take a look at goals and shots (shots taken and shots on target) per match played.

No real surprises on the the number of goals scored per match front. Man U, Arsenal, and Chelsea are right at the top at 2-2.5 goals per match, and Wigan, West Ham, and Fulham bring up the rear at 1 or less (Wigan is positively woeful). Perhaps slightly surprising is that Bolton and Chelsea are virtually tied in the goals per match category around 2 goals, and it's testament to Bolton's great run so far this year.

The top teams also shoot more on average, and they tend to be more accurate when they do shoot. So Arsenal's and Chelsea's 16+ shots per match, with over half of them on target, make them fierce offensive machines (with Tottenham not far behind). Then there's a large group of teams in the middle range, shooting about 12-14 times per match, with about 6-7 shots on target. Finally, there are - perhaps surprisingly for Birmingham fans - Birmingham City and Blackburn who take only about 9 shots on goal that find the target about half the time as well.

We can use these data to compute goal to shot ratios, as well as accuracy and conversion ratios (shots on target as a proportion of total shots as well as goals as a proportion of shots on target, respectively). Here's what you see - split up by whether teams played home or away.